Retention and bias assessment among an online trial of men who have sex with men in Shandong province, China (Preprint)

Abstract

In prospective studies, the rate of loss of follow-up (LTFU) has a significant impact on the validity of the results. Although some studies have explored the influencing factors of retention in MSM population, most of them are concentrated in a single region and mostly offline research, and there is no more understanding of the predictors retained in online prospective studies of MSM population.

To investigate the retention in an online trial of men who have sex with men (MSM) in 10 cities of Shandong province and evaluate the effect of bias on the relationship between exposures and HIV testing.

We selected 10 cities in Shandong province as the study area to carry out a randomized controlled trial and followed up for about 12 months. Participants were followed up approximately every three months for a total of four times. Information collected included social demographic characteristics, sexual behavior, HIV testing, etc. LTFU was defined as the respondents who met the inclusion criteria but had not participated in any follow-up visits after entering the cohort. LTFU and the number of follow-up visits, which were used as the indicators to evaluate the retention of the respondents, were estimated by Cox proportional hazard model and a generalized Poisson regression model, respectively. In addition, the bias was assessed to further explain the impact of LTFU.

We recruited 935 participants in the study, 184 of whom did not participate in any follow-up, and the LTFU rate was 19.68%. The results of multivariate analysis showed that both the LTFU and the number of follow-up visits were closely related to the city and sexual orientation. LTFU and less follow-up visits were common among heterosexual/bisexual participants, and relatively less in homosexual participants. Participants in cities with lower GDP were more likely to be lost and had less follow-up visits than those in cities with higher GDP. Besides, the higher the educational level, the more times to participate in follow-up visits. The results of bias assessment suggested that the bias caused by LTFU underestimated the relationship between cities and HIV testing, but overestimated the relationship between age, sexual orientation disclosure and HIV testing.

The retention of an online trial among MSM in 10 cities of Shandong province was not ideal and LTFU mainly occurred in the first follow-up. Future strategies should focus on cities with lower GDP, sexual orientation (heterosexual/bisexual) and less educated participants to improve the retention rate of studies. Moreover, bias assessment can be used to evaluate the association of exposure and outcome, which made the interpretation of results more credible.